Abstract
BACKGROUND: A large body of evidence supports a link between type 2 diabetes mellitus (T2DM) and cognitive function, including dementia. However, longitudinal studies on the association between T2DM and decline of cognitive function are scarce and reported mixed results, and we hence set out to investigate the cross-sectional and longitudinal association between T2DM and global as well as domain-specific cognitive performance. METHODS: We used multivariable regression models to assess associations of T2DM with cognitive performance and cognitive decline in a subsample of a population-based prospective cohort study (ESTHER). This subsample (n = 732) was aged 70 years and older and had participated in telephone-based cognitive function assessment (COGTEL) measuring global and domain-specific cognitive performance during the 5- and 8-year follow-up. RESULTS: Total COGTEL scores of patients with prevalent T2DM were 27.4 ± 8.3 and 29.4 ± 8.7 at the 5- and 8-year measurements, respectively, and were roughly two points lower than those of T2DM-free participants after adjustment for age and sex. In cross-sectional models, after adjustment for several potential confounders, performance in verbal short-term and long-term memory tasks was statistically significantly lower in participants with T2DM, but the association was attenuated after further adjustment for vascular risk factors. The difference in total COGTEL scores reflecting global cognitive function by T2DM status after full adjustment for confounders and vascular risk factors was equivalent to a decrement in global cognitive function associated with a four-year age difference. In longitudinal models, a statistically significantly stronger cognitive decline in patients with T2DM was observed for working memory. CONCLUSIONS: In this sample of older individuals, T2DM was associated with worse performance and stronger decline in a cognitive function test. Memory-related domains were found to be particularly sensitive to T2DM. Further large-scale prospective studies are needed to clarify potential T2DM-related predictors of cognitive decline and possible consequences on the abilities to perform patient self-management tasks in diabetes care.